— It’s a debate as old as time, or at least as old as the first conference tournament: Does winning a conference tournament help or hurt a team’s NCAA tournament prospects?
The benefits of winning a league tournament are obvious. The extra wins can improve a team’s seeding, the extra games can provide more practice against tough competition, and the extra confidence from lifting one trophy can help propel a team toward lifting another.
But the doubters also have reasonable arguments. Playing on several consecutive nights can drain a team. And some claim that a “good loss” can wake a team up and light a fire in the players’ bellies. (I’d be a little freaked out if I woke up to find my stomach on fire, but that’s just me.)
So which camp is correct? Mike Rutherford of Card Chronicle tackled this subject recently, arguing that winning your tournament is good. I’d like to back him up by going back a few more years, and taking a slightly more comprehensive approach.
To start, let’s ask a simple question. How far in the NCAA tournament do teams generally advance after winning their conference tournament? How about after making it to the conference final, but losing, since these teams should still suffer from tired legs syndrome? And after losing earlier than the final round?
This chart shows, for each of the above groups, the percentage of teams that survived until each round of the NCAA tournament. It only includes major conference teams from 1997-1998 to 2010, and excludes some old Pac-10 teams, from before the Pac-10 tournament existed.
Look at the green line. It shows that, for example, about 90 percent of the teams that won their conference tournament survived until Round 2 of the NCAA tournament, and about 10 percent won the national title.
There’s an obvious trend there. The teams that won their conference tournaments did better in the NCAA tournament, by a large margin.
Ah, but I’m clumping all the seeds together here. What happens if we split them into groups of high and low seeds? I counted 1 through 6 seeds as “high.” Anything lower and a team is already likely a huge underdog in one of its first two games.
Same as above, but for high seeds:
And for low seeds:
Now we see that all of the performance edge for the conference tournament-winning teams is due to the success of the higher seeds. For teams in the middle seed lines, conference tournament results seem to mean little.
That settles it, right? Though it means nothing to the mid-tier teams, for teams in the high seed lines, winning their conference tournament is a great sign of future success.
Not exactly. The pattern we’re seeing probably isn’t due to some magical quality bestowed on conference champions, but to the fact that the teams that win conference tourneys tend to be very good, and those very good teams face relatively easier opponents due to their high seeds. If we account for those two facts, the conference tournament boost may disappear.
To test this idea, I used a technique called logistic regression. What that boils down to is giving my statistics program a few different pieces of information about each team, and then letting it figure out which ones are the most important in explaining a team’s performance.
I used three things as inputs to the regression program:
That the ratings are from the start of the NCAA tournament is important; end-of-year ratings would skew the results because the NCAA tournament games themselves would already be factored into the ratings.
For past major conference teams seeded 1 through 6, I had the regression program predict two different probabilities: the chance of a team making the Sweet 16 (to test the “tired legs” protest), and the chance of a team advancing to the Final Four (to represent a deep tourney run).
The regression program spits out a number called a “z value” that indicates whether a variable is statistically significant in predicting the NCAA performance of a team. Values over 2 are usually considered significant, and anything over 3 is considered extremely significant. Numbers near zero mean the variable could just as easily be a random number in the eyes of the program. I bet you can guess the results:
Once you account for team rating and seed, conference tournament success isn’t a predictor of NCAA tournament results. If that’s the case, why did we see the earlier result, showing that high seeds who win their conference tournament do well in the Big Dance?
Quite simply, it’s because the best teams tend to win the conference tournaments. This chart shows the percentage of teams on each seed line that won their tournament. For example, 64 percent of No. 1 seeds won their conference title.
Because the “won conference tournament” group had far more top seeds than the other groups, its performance was bound to be better.
Of course, it’s undoubtedly true winning the conference tournament helps a team gain a higher seed. In fact, I expected that between two teams of equal quality, the one with a conference tournament win would be seeded higher. But using the same logistic regression technique I described above, I found that wasn’t the case. The committee has little to no bias toward tournament winners.
Over the next few days, people may try to convince you that winning a conference tournament is a good omen, or that bowing out early keeps a team fresh for the main event. Don’t listen to either. The latter claim has no supporting evidence, and the former is true only because winning a conference tournament correlates with the most important factor in NCAA tournament success: being a good team.